Applicability of published experimental works as a knowledge source in recommendation of gold ore processing workflows

Abstract

The experimental work is the most time consuming and expensive part of the process design. A case-based reasoning (CBR) methodology can be used to assist in the process design. Auric Adviser is a CBR system under development for recommendation of gold ore processing workflows. In Auric Adviser the knowledge in gold ore processing is represented in two models. The first model holds the knowledge needed to recommend process chains already used in industry. The second knowledge model is intended to recommend the most efficient unit processes based on research results. The objective of this study was to analyse the information richness of journal articles and other publications concerning single process steps of gold extraction. The aim was to study the applicability of these publications as a source for the second Auric Adviser knowledge model. In this study, 25 publications concerning leaching of gold were analysed and information was extracted in a case base. The case base was taken as either a process or experimental description with clearly defined differences to other descriptions. In total, 218 cases that described results of gold leaching were extracted from the sources. The analysis of descriptions showed that the knowledge necessary for design the second Auric Adviser model can be elicited from journal articles and other publications concerning single process steps. The trends in the case description were that the gold content and process outcome were usually well described. Nevertheless, the information richness varied in descriptions of raw materials, experimental arrangements but the results were often missing details. The incompleteness of information causes challenges in the process comparison although it does not prevent the CBR system to work.